lambda_handler.py

import os, subprocess, tempfile, base64, time, sys

SCRIPT_DIR = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'tesseract-lambda')
LIB_DIR = os.path.join(SCRIPT_DIR, 'lib')

def ocr(event, context):
    with tempfile.NamedTemporaryFile() as temp:
        temp.write(base64.b64decode(event['data']))
        temp.flush()

        ocr_name = temp.name
	convert_time = 0
        if event['filename'].split('.')[1] == 'pdf':
            ocr_name += '.tiff'
            cmd = 'gs -dNOPAUSE -r300x300 -sDEVICE=tiffg4 -dBATCH -sOutputFile={} {}'.format(
                ocr_name,
                temp.name
            )
            print cmd
            try:
        	start = time.time()
                output = subprocess.check_output(cmd, shell=True)
		convert_time = time.time() - start
                print output
            except subprocess.CalledProcessError as convertE:
                print convertE.output
                raise convertE


        command = 'LD_LIBRARY_PATH={} TESSDATA_PREFIX={} {}/tesseract {} {}'.format(
            LIB_DIR,
            SCRIPT_DIR,
            SCRIPT_DIR,
            ocr_name,
            temp.name,
        )

        try:
	    start = time.time()
            output = subprocess.check_output(command, shell=True)
            ocr_time = time.time() - start
        except subprocess.CalledProcessError as ocrE:
            print ocrE.output
            raise ocrE

        with open(temp.name+'.txt', 'r+') as outfd:
            ocr = base64.b64encode(outfd.read())

        os.remove(temp.name+'.txt')

        ret_name = event['filename'].split('.')[0] + '.txt'

        return {'data':ocr, 'filename':ret_name, 'ocr_time':ocr_time, 'convert_time':convert_time}

def lambda_handler(event, context):
    fn = {
        'ocr': ocr
    }[event['op']]

    return fn(event, context)